U.S. patent number 10,482,918 [Application Number 16/104,332] was granted by the patent office on 2019-11-19 for changing bit spacing for selected symbols written to a magnetic recording medium.
This patent grant is currently assigned to Seagate Technology LLC. The grantee listed for this patent is Seagate Technology LLC. Invention is credited to Mehmet Fatih Erden, Steven Douglas Granz, Stephanie Hernandez.
![](/patent/grant/10482918/US10482918-20191119-D00000.png)
![](/patent/grant/10482918/US10482918-20191119-D00001.png)
![](/patent/grant/10482918/US10482918-20191119-D00002.png)
![](/patent/grant/10482918/US10482918-20191119-D00003.png)
![](/patent/grant/10482918/US10482918-20191119-D00004.png)
![](/patent/grant/10482918/US10482918-20191119-D00005.png)
![](/patent/grant/10482918/US10482918-20191119-D00006.png)
![](/patent/grant/10482918/US10482918-20191119-D00007.png)
United States Patent |
10,482,918 |
Erden , et al. |
November 19, 2019 |
Changing bit spacing for selected symbols written to a magnetic
recording medium
Abstract
A set of patterns written to a magnetic recording medium are
found that result in a relatively high and/or low error when read
back. Upon writing subsequent to the determining of the set of
patterns, one pattern of the set of the patterns is identified
within a data segment ready to be written to the magnetic recording
medium. The data segment is written with a changed bit spacing in
response to identifying that the one pattern of the set of the
patterns is within the data segment.
Inventors: |
Erden; Mehmet Fatih (St. Louis
Park, MN), Hernandez; Stephanie (Plymouth, MN), Granz;
Steven Douglas (Shakopee, MN) |
Applicant: |
Name |
City |
State |
Country |
Type |
Seagate Technology LLC |
Cupertino |
CA |
US |
|
|
Assignee: |
Seagate Technology LLC
(Cupertino, CA)
|
Family
ID: |
68536431 |
Appl.
No.: |
16/104,332 |
Filed: |
August 17, 2018 |
Current U.S.
Class: |
1/1 |
Current CPC
Class: |
G11B
20/1217 (20130101); G11B 20/10009 (20130101); G11B
20/182 (20130101); G11B 20/105 (20130101) |
Current International
Class: |
G11B
5/09 (20060101); G11B 20/18 (20060101); G11B
20/10 (20060101) |
References Cited
[Referenced By]
U.S. Patent Documents
Primary Examiner: Agustin; Peter Vincent
Attorney, Agent or Firm: Mueting, Raasch & Gebhardt,
P.A.
Claims
What is claimed is:
1. A method, comprising: writing a plurality of different symbols
to a magnetic recording medium at a nominal bit timing; determining
a first set of the symbols that result in a relatively high error
when read back; upon writing subsequent to the determining of the
first set of symbols, identifying one symbol of the first, set of
symbols within a data segment ready to be mitten to the magnetic
recording medium; and writing the data segment with increased bit
spacing relative to the nominal bit timing in response to
identifying that the one symbol of the first set of symbols is
within the data segment.
2. The method of claim 1, further comprising: determining a second
set of the symbols written to the magnetic recording medium that
result in a relatively low error when read back; upon writing
subsequent to the determining of the second set of symbols,
identifying one symbol of the second set of the symbols within a
second data segment ready to be written to the magnetic recording
medium; and decreasing a second bit spacing of the second data
segment relative to the nominal bit timing in response to
identifying the one symbol of the second set of the symbols within
the data segment.
3. The method of claim 1, wherein determining the first set of
symbols mitten to a magnetic recording medium that result in a
relatively high error comprises writing pseudo-random bit sequences
and reading back the pseudo-random bit sequences to determine the
set of symbols.
4. The method of claim 3, wherein the first set of symbols have a
lowest signal-to-noise ratio of all symbols in the pseudo-random
bit sequences.
5. The method of claim 3, wherein the first set of symbols have a
highest bit error rate of all symbols in the pseudo-random bit
sequences.
6. The method of claim 1, wherein determining the first set of
symbols written to a magnetic recording medium that result in a
relatively high error comprises reading back previously written
user data to determine the set of symbols.
7. The method of claim 1; wherein the bit spacing is increased by a
fractional amount between 1% to 100% of a single bit spacing.
8. A method, comprising: writing a plurality of different symbols
to a magnetic recording medium at a nominal bit timing; determining
a set of the symbols that result in a relatively low error when
read back; upon writing subsequent to the determining of the set of
symbols, identify one symbol of the set of the symbols within a
data segment ready to be written to the magnetic recording medium;
and write the data segment with decreased bit spacing relative to
the nominal bit timing in response to identifying that the one
symbol of the set of the symbol is within the data segment.
9. The method of claim 8, wherein determining the set of symbols
written to a magnetic recording medium that result in a relatively
low error comprises writing pseudo-random bit sequences and reading
back the pseudo-random bit sequences to determine the set of
symbols.
10. The method of claim 9, wherein the set of symbols have a
highest signal-to-noise ratio of all symbols in the pseudo-random
bit sequences.
11. The method of claim 9, wherein the set of symbols have a lowest
bit error rate of all symbols in the pseudo-random bit
sequences.
12. The method of claim 8, wherein determining the set of symbols
written to a magnetic recording medium that result in a relatively
low error comprises reading back previously written user data to
determine the set of symbols.
13. The method of claim 8, wherein the bit spacing is decreased by
a fractional amount between 1% to 50% of a single bit spacing.
14. An apparatus, comprising: interface circuitry coupled to a head
that reads from and writes to a magnetic recording medium; and a
controller coupled to the interface circuitry and operable to:
determine a set of symbols written to the magnetic recording medium
that result in one of a relatively high error or relatively low
error when read back; upon writing subsequent to the determining of
the set of symbols, identify one symbol of the set of the symbols
within a data segment ready to be written to the magnetic recording
medium; and write the data segment with a changed bit spacing in
response to identifying that the one symbol of the set of the
symbols is within the data segment.
15. The apparatus of claim 14, wherein determining the set of
symbols written to a magnetic recording medium that result in the
relatively high or low error comprises writing pseudo-random bit
sequences and reading back the pseudo-random bit sequences to
determine the set of symbols.
16. The apparatus of claim 15, wherein the set of symbols have a
lowest or highest signal-to-noise ratio of all symbols in the
pseudo-random bit sequences.
17. The apparatus of claim 15, wherein the set of symbols have a
highest or lowest bit error rate of all symbols in the
pseudo-random bit sequences.
18. The apparatus of claim 14, wherein determining the set of
symbols written to a magnetic recording medium that result in the
relatively high or low error comprises reading back previously
written user data to determine the set of symbols.
19. The apparatus of claim 14, wherein the set of symbols results
in the relatively high error, the bit spacing being increased by a
fractional amount between 1% to 100% of a single bit spacing.
20. The apparatus of claim 14, wherein the set of symbols results
in the relatively low error, the bit spacings being decreased by a
fractional amount between 1% to 50% of a single bit spacing.
Description
SUMMARY
The present disclosure is directed to changing bit spacing for
selected patterns written to a magnetic recording medium. In one
embodiment, a set of patterns written to a magnetic recording
medium are found that result in a relatively high and/or low error
when read back. Upon writing subsequent to the determining of the
set of patterns, one pattern of the set of the patterns is
identified within a data segment ready to be written to the
magnetic recording medium. The data segment is written with a
changed bit spacing in response to identifying that the one pattern
of the set of the patterns is within the data segment.
These and other features and aspects of various embodiments may be
understood in view of the following detailed discussion and
accompanying drawings.
BRIEF DESCRIPTION OF THE DRAWINGS
The discussion below makes reference to the following figures,
wherein the same reference number may be used to identify the
similar/same component in multiple figures.
FIG. 1 is a block diagram illustrating a read head and recording
medium according to an example embodiment;
FIG. 2 is a graph showing errors of different patterns according to
an example embodiment;
FIGS. 3-5 are flowcharts of a process according to an example
embodiment;
FIG. 6 is a diagram showing a signal waveform according to an
example embodiment;
FIGS. 7-9 are flowcharts of a process according to another example
embodiment;
FIG. 10 is a diagram showing a signal waveform according to another
example embodiment;
FIG. 11 is a block diagram of a system and apparatus according to
an example embodiment; and
FIG. 12 is a graph showing results of a simulation according to an
example embodiment.
DETAILED DESCRIPTION
The present disclosure generally relates to data storage devices
that utilize magnetic storage media, e.g., disks. These data
storage devices utilize write transducers (e.g., a magnetic coil)
that apply a changing magnetic field to the recording medium. The
applied field changes magnetic orientation in regions of the
recording medium, which define bit boundaries of the stored data. A
series of adjacent bits recorded along a circular path defines a
data track on the recording medium. A read transducer (e.g.,
magnetoresistive sensor) can later traverse the track and detect
the magnetic transitions. These transitions form a signal via the
read transducer which is decoded to recover the stored data.
During the write process, user bits are first encoded and then
written onto the magnetic recording medium using a write-head.
During the read process, read head detects the encoded data from
the desired location on the medium, and the noisy analog signals
are then processed by the read channel architectures to extract the
written data as a series of bits that represent the recorded data.
These bits are then decoded to extract the user data, which may be
further processed (e.g., error correction codes applied) before
being communicated to a host computer.
For a given system operating point, there are bit patterns written
to the recording medium that result in worse resolution than other
patterns during reading. These bit patterns result in worse
detection and correction capability for read-channel architectures.
The performance of the overall system is mainly defined with those
patterns as they play a role in design of the channel codes that
correct system errors. On the other hand, the patterns with excess
resolution don't affect the system performance that much, yielding
opportunities to further optimize the system. In this disclosure,
methods and apparatuses are proposed which change the resolution of
bit patterns written on magnetic media in order to increase the
overall system capacity.
In FIG. 1, a block diagram illustrates a magnetic head 104 and
recording medium (e.g., magnetic disk 102) according to an example
embodiment. Differently shaded regions 100 represent bits recorded
on the disk 102. The disk 102 rotates (as indicated by arrow 103)
while a magnetic head 104 (also referred to as a recording head,
read/write head, read head, write head, slider, etc.) is held over
a top surface of the medium 102. The magnetic head 104 can be moved
over different tracks of the disk 102 by an arm 110.
The magnetic head 104 has a read transducer 106 (also referred to
here as a reader) that produces an analog data signal 108 in
response to the changes in magnetic orientation of the recorded
bits. Each positive or negative transition of the signal 108
corresponds to a bit transition. Note that the form of signal 108
is provided for purposes of illustration, and may take other forms
depending on the signal path, encoding/decoding schemes, etc. The
bit transitions detected in signal 108 are detected and time
adjusted so as to correspond to a transition of a data clock. The
data clock signal may be implied, e.g., timing of the signal 108
may be derived based on the signal itself instead of referencing a
hardware clock or other time source.
In some cases, the read channel that processes the signal 108 may
experience greater than average errors in detecting and decoding
particular patterns of user data bits that are encoded in the
signal 108. These errors may be caused by a number of phenomena,
such as less signal and more noise associated with that pattern,
non-linear response of the recording medium 102 and/or reader 106,
characteristics of the electrical pathways that distort the signal
108 before it is received by the processing circuitry, etc. A read
channel (not shown) that processes the signal 108 may experience a
higher amount of errors when detecting and decoding these
particular patterns. Similarly, other patterns may be enhanced by
the above-described phenomena, and therefore may exhibit lower than
average errors when being detected and decoded.
In embodiments described below, a drive is configured to analyze
and classify patterns based on errors exhibited when reading the
patterns back. A write channel is configured to increase the
resolution of worst case patterns, and alternatively or
additionally, reduce the resolution of the patterns having smaller
than average errors. In FIG. 2, a graph 200 shows how patterns may
be categorized to perform one or both of these
resolution-adjustment operations according to an example
embodiment.
The graph in FIG. 2 is a simplified example that assumes the
bit-error rate (BER) of N-random symbols stored on drive have been
averaged over a large sample size. The symbols correspond to
pattern of magnetic fields written to the recording medium and then
read back, which does not necessarily correspond to the user data
decoded from the values. Each of the bars in the graph 200
represent a change in BER over average BER for each of the symbols.
The symbols 201 represent the five patterns having the highest BER
over average and the symbols 202 represent the five patterns having
the lowest BER over the average. It will be understood that this
graph is a simplified example, and different pattern sizes and
maximum/minimum values may be selected in other embodiments. In
addition, other characteristics such as signal-to-noise ratio may
be used instead of BER to select the symbol patterns 201, 202.
In FIG. 3, a flowchart shows a procedure used to improve resolution
of patterns with high errors (e.g., symbols 201 in FIG. 2)
according to an example embodiment. The procedure involves
determining 300 a set of patterns written to a magnetic recording
medium that result in a relatively high error when read back. Upon
writing subsequent to the determining 300 of the set of patterns,
one pattern of the set of the patterns is identified 301 within a
data segment ready to be written to the magnetic recording medium.
In response to identifying 301 that the one pattern of the set of
the patterns is within the data segment, the data segment is
written 302 with increased bit spacing.
In FIGS. 4 and 5, flowcharts show a more detailed procedure to
improve resolution according to an example embodiment. A pre-write
procedure 400 shown in FIG. 4 involves forming 401 look-up table
405 (or other data structure) having patterns with bad resolution.
For example, this table may be formed by looking the error events
at the output of the channel detector with highest percentage of
occurrences. For example, this may include the top-N error events,
or error events with occurrence percentages higher than a
predefined value. Within those error events, patterns resulting
into bad system resolution are identified 402 and added 403 to the
look-up table 405.
A write process 500 is shown in FIG. 5 that involves the writing of
an individual sector, although may be used to write any contiguous
data regions. In the write process 500, a window of size L is
formed 501, where L is the length of the longest pattern in the
look-up table 405. The first data segment is loaded 502 (e.g., into
a memory buffer) and window is slid 503 along the data segment
starting from the first channel bit until one of the patterns in
the table 405 is detected 504. If one of the patterns is detected
(block 504 returns `yes`), write channel pre-compensation circuitry
is set such that the spacing between each transition within the
pattern is increased 505 by a predefined fractional amount of the
bit width, .delta..sub.1. The segment is written 506 and it is
determined 507 whether the end of sector has been reached. If not,
then the next segment is loaded 508 and the process repeats until
the end of the data sector.
In a drive that uses heat-assisted magnetic recording (HAMR), a
laser or other heat source forms a hot spot on the disk while
recording. This locally lowers the coercivity at the hotspot
allowing a write pole to change the magnetic orientation of the
hotspot without affecting areas just outside the hotspot. In case
of HAMR, in addition to utilizing the pre-compensation circuitry,
laser power modulation can have a similar effect. Because an
increase or decrease in laser current changes a size of the
hotspot, it will also change locations of bit transitions if timed
appropriately with the bit transitions.
In FIG. 6, a signal diagram shows the increasing of bit resolution
according to an example embodiment. Waveform 600 represents nominal
bit timing used to write a series of bits to the recording medium.
Waveform 602 (dashed lines) represents the same series of bits
written with the time of each transition increased by the
fractional amount .delta..sub.1. Assuming this increase is applied
to each transition, the total length of the data sector will
increase. In some applications, such as cold storage, the sector
boundaries can be dynamic and so this will not result in
overwriting subsequent sectors. In other applications, the
sector-to-sector spacing can be overprovisioned to account for some
amount of sector size change. Note that upon reading the signal 602
back, the timing recovery circuits in the read channel will adjust
to compensate for the increased spacing between signal transitions
when detecting the bits.
One specific case of the proposed algorithm above can be visualized
as a generalization of the conventional (d, k) modulation codes
with fractional d parameter. A (d, k) modulation code corresponds
to having d repetitions of the bit when a transition is observed
and having a transition after k runs of constant bits. For example,
(0,7) modulation code does not require any forced repetition when a
transition happens, while forces a transition when 7 consecutive 1s
or -1s are observed. On the other hand, (1,7) code, where d=1,
requires 1 forced repetition of the same bit when a transition
occurs while the k constraint acts the same. More specifically, for
a sequence of channel bits equal to 1 1 -1 1 1 1 1 1 1 1 1 1, a
(0,7) encoder yields 1 1 -1 1 1 1 1 1 1 1 -1 1 1 while (1,7)
encoder yields 1 1 -1 -1 1 1 1 1 1 1 1 -1 -1 1 1.
Having a d constraint in a (d, k) code introduces redundancies.
Even when k is approaching infinity, the capacity of the (1, k)
code is 69.42%, results in more than 30% redundant bits, which
becomes inefficient for magnetic recording. On the other hand, as
defined above, the proposed algorithm (for the pattern of having a
transition) corresponds to repeating the bit by a fractional amount
of it whenever there are two consecutive transitions (not when
there is only one single transition as defined conventionally). For
example, increasing the bit length by 10% whenever there are two
consecutive transitions corresponds to d equal to 0.1 in this
terminology. The extra length required on the magnetic medium with
fractional d parameter requires around 3% extra code redundancy and
helps to better optimize the system by choosing the fraction
accordingly. In embodiments described below, the bit transitions
can be increased between by a fractional amount between 1% to 100%
of a single bit spacing. In embodiments where the bit transitions
are decreased, the decrease would be by a fractional amount between
1% to 50% of a single bit spacing.
As noted above, a similar process can be applied to patterns that
have unusually low error rates, except that bit resolution is
decreased in such a case instead of increased. In FIG. 7, a
flowchart illustrates a procedure used to reduce resolution of
patterns with low errors (e.g., symbols 202 in FIG. 2) according to
an example embodiment. The procedure involves determining 700 a set
of patterns written to a magnetic recording medium that result in a
relatively low error when read back. Upon writing subsequent to the
determining 700 of the set of patterns, one pattern of the set of
the patterns is identified 701 within a data segment ready to be
written to the magnetic recording medium. In response to
identifying 701 that the one pattern of the set of the patterns is
within the data segment, the data segment is written 702 with
decreased bit spacing.
In FIGS. 8 and 9 flowcharts show a more detailed procedure to
change resolution according to an example embodiment. In FIG. 8, a
prewrite procedure 800 involves forming 801 a look-up table 805
used to store patterns with good resolution. The patterns are
identified 802 and added to the table 805. For example, identifying
802 the patterns may involve writing a pseudo-random bit sequence
(PRBS) to the recording medium and the reading the PRBS read back.
The signal-to-noise ratio (SNR) of each pattern in the PRBS is
calculated and the patterns with good (high) SNRs are identified
802 and added 803 to the table 805.
In FIG. 9, a write process 900 reduce the resolution of the
patterns with good resolution within a data sector (or other data
region). In the write process 900, a window of size L is formed
901, where L is the length of the longest pattern in the look-up
table 805. The first data segment is loaded 902 (e.g., into a
memory buffer) and window is slid 903 along the data segment
starting from the first channel bit until one of the patterns in
the table 805 is detected 904. If one of the patterns is detected
(block 904 returns `yes`), write channel pre-compensation circuitry
is set such that the spacing between each transition within the
pattern is decreased 905 by a predefined fractional amount of the
bit width, .delta..sub.2. The segment is written 906 and it is
determined 907 whether the end of sector has been reached. If not,
then the next segment is loaded 908 and the process repeats until
the end of the data sector.
In FIG. 10, a signal diagram shows the decreasing of bit resolution
according to an example embodiment. Waveform 1000 represents
nominal bit timing used to write a series of bits to the recording
medium. Waveform 1002 (dashed lines) represents the same series of
bits written with the time of each transition decreased by the
fractional amount 62. Assuming this decrease is applied to each
transition, the total length of the data sector will decrease. As
with other embodiments, upon reading the signal 1002 back, the
timing recovery circuits in the read channel will adjust to
compensate for the decreased spacing between signal transitions
when detecting the bits.
In FIG. 11, a block diagram illustrates a system and apparatus
according to an example embodiment. A processor/controller 1101
include a general-purpose and/or special purpose processors, logic
circuits, signal processors, etc., that control a write channel
1104 and a read channel 1106. The write channel encodes data into
signals that are sent to one or more write transducers 1108 that
write to a magnetic recording medium 1112. The read channel 1106
receives signals from one or more read transducers 1110 that are
held over the recording medium 1112.
The read channel 1106 performs a pattern identification function
1116 that looks at a set of patterns (e.g., byte patterns,
multi-byte patterns, etc.) read back from the recording medium
1112. These patterns are analyzed to determine whether an unusually
high or low error is associated with the patterns. The error may be
estimated by looking at any combination of BER and SNR, in addition
to other indicators, e.g., dibit response. Generally, a low SNR or
high BER indicates high error rate, and high SNR or low BER
indicates low error rate. This estimation may be performed by
reading user data, PRBS, and/or a non-random sequence of test
patterns. Patterns that result in unusually high or low measures of
readback error are stored in a pattern database 1118.
After patterns have been identified and stored in the pattern
database 1118, the write channel 1104 performs a bit resolution
adjustment function 1114 that changes a timing of magnetic
transitions written to the recoding medium 1112 by increasing the
bit transition timing (thus the resolution) of error-prone patterns
or by decreasing the bit transition timing of patterns that exhibit
little error. Note that the pattern identification function 1116
and bit resolution adjustment function 1114 can be continuously
operating during the life of the drive, such that the identified
patterns and amount of bit resolution adjustment can change and
adapt based on changing operating conditions (e.g., temperature,
vibration, electromagnetic interference), aging of the drive,
etc.
In order to estimate the effect of increasing resolution on drive
performance, a modeling simulation was run. This simulation
involved determining SNR for different pure tone frequencies from 1
T to 6 T using a HAMR drive. Single tone isolated tracks were
written on AC erased media and then read-back. The read-back
waveforms were captured using a micromagnetic reader and SNR
measured for multiple passes. The SNR measured as a function of
tone length is shown in the graph of FIG. 12. From 1 T to 2 T, SNR
increases by 1.7 dB per 0.1 T. From 2 T to 5 T, SNR increases by 2
dB per 0.5 T (or about 0.4 dB per 0.1 T). Beyond 5 T, less increase
in SNR is observed. Given that performance is limited by the
shortest marks, this suggests that much SNR gain be achieved if the
short marks (1 T, 2 T) are increased in length, while decreasing
the length of the long marks.
Unless otherwise indicated, all numbers expressing feature sizes,
amounts, and physical properties used in the specification and
claims are to be understood as being modified in all instances by
the term "about." Accordingly, unless indicated to the contrary,
the numerical parameters set forth in the foregoing specification
and attached claims are approximations that can vary depending upon
the desired properties sought to be obtained by those skilled in
the art utilizing the teachings disclosed herein.
The various embodiments described above may be implemented using
circuitry, firmware, and/or software modules that interact to
provide particular results. One of skill in the arts can readily
implement such described functionality, either at a modular level
or as a whole, using knowledge generally known in the art. For
example, the flowcharts and control diagrams illustrated herein may
be used to create computer-readable instructions/code for execution
by a processor. Such instructions may be stored on a non-transitory
computer-readable medium and transferred to the processor for
execution as is known in the art. The structures and procedures
shown above are only a representative example of embodiments that
can be used to provide the functions described hereinabove.
The foregoing description of the example embodiments has been
presented for the purposes of illustration and description. It is
not intended to be exhaustive or to limit the embodiments to the
precise form disclosed. Many modifications and variations are
possible in light of the above teaching. Any or all features of the
disclosed embodiments can be applied individually or in any
combination are not meant to be limiting, but purely illustrative.
It is intended that the scope of the invention be limited not with
this detailed description, but rather determined by the claims
appended hereto.
* * * * *